Neuroanatomical profile of externalizing problems level in adults: a machine learning study

Externalizing problems (EXT) refers to negative behaviors directed externally into the environment. Though a growing literature on structural magnetic resonance imaging (sMRI) research has supported this concept by revealing neurobiological profiles underlying EXT, it is still unclear whether previously identified relationships are generalizable (out-of-sample) and how cortical thickness and cortical surface area contribute to EXT. In this study, using the Human Connectome Project Young Adult dataset (N = 1098), a machine learning cross-validated elastic net regression approach was conducted to characterize the neuroanatomical pattern of sMRI variables associated with EXT. The results revealed a multi-region neuroanatomical signature predicted EXT and this was robust in a held-out test set (morphometry-only R²= 2.69%, morphometry + demographics R² = 6.45%). The neuroanatomical pattern included regions implicated in the frontal-limbic circuit. The relationship of these regions with EXT was further supported by univariate linear mixed effects modeling results. Taken together, these findings provided evidence that a machine learning-derived neuroanatomical profile encompassing various theoretically relevant brain circuits could predict EXT in a large sample of healthy adults, which may inform monitoring and prevention strategies for the broad externalizing dimension in community populations.